The Mechanic’s Ear

IN 2002, THE PHILOSOPHER MATTHEW CRAWFORD left a Washington think tank to open a motorcycle repair shop in Richmond, Virginia. From the outside, it looked like a move away from intellectual work. Crawford saw it differently.

Repairing motorcycles required knowledge that no manual could fully contain. The experienced mechanic might know the specification, the fault codes and the recommended sequence of repair. But they also knew what a failing bearing sounded like, how an engine felt when it was running badly and when the apparent fault was only a symptom of something deeper.

His 2009 book, Shop Class as Soulcraft, drew on that journey to describe how judgement is learned. It showed how some knowledge can be written down, standardised and taught, whilst some must be acquired through practice, because it lives in the relationship between the person, the task and the material in front of them. And how this can often take years to accumulate and can disappear remarkably quickly when the people carrying it leave.

Crawford called this embodied knowledge. Philosophy more often calls it tacit knowledge. Construction tends not to call it anything at all. It hopes it is there. Now, as the UK redesigns its construction training system, it is deciding not only how many people to train, but what forms of knowledge the next system must preserve.

The wrong question

The demographic picture is familiar enough to summarise in a line: employment at its lowest share of the UK workforce in over a century, more people leaving each year than joining, almost half of apprentices dropping out before completing. The question is no longer whether the training system needs reform. It does. The question is what it is training people for.

If construction’s production model is evolving, and it is, increasingly toward factory manufacture, digital coordination, and automated site assembly, then training people into the existing model feeds them into a system that is itself in transition. Training people for the present without preparing them for where the future construction sector is heading stores up a second crisis inside the first.

This is where the workforce debate keeps slipping. It treats the problem as one of supply: how many people, through which pathways, funded by what mechanism. Those questions matter. But they assume the destination is stable. It is not.

The harder question, the one Crawford forces, is whether the system being built can hold the wisdom that matters.

Knowledge is knowing a tomato is a fruit. Wisdom is knowing not to put it in a fruit salad.

What a drawing cannot say

Consider for example, a congested beam-column detail on an RC frame. The structural drawing shows the reinforcement layout; the model shows no clashes. But the experienced steel fixer knows the physical sequence: which bars must go in first, which can be threaded through afterwards, where to leave windows for the poker vibrator, where a lap splice will make the next bar physically impossible to place. A T40 bar weighs nearly ten kg per metre. You do not reposition it once the cage is built around it.

The mechanical installer who has worked dozens of ceiling voids can look at a services coordination drawing and tell you which fire damper will be inaccessible in eighteen months, before the ceiling tiles are ordered. Not because the drawing is wrong, but because she knows what happens to access once the cable trays, containment, and lagging go in.

This is not edge-case knowledge. It is the ordinary texture of how buildings get built. And much of it resists easy codification. It is not that these judgements can never be captured in a system. It is that they emerge through situated practice and accumulate over time. The transition now under way must find ways to retain and transfer them before the people carrying them leave.

The ears that designed the robots

Japan confronted the same problem earlier and more acutely. When a hard cap on overtime hours exposed the workforce decline that excessive working had been masking, the response was not simply to train more people. New occupational categories were created. National training pathways were built around them. And the experienced craft workforce was involved in the design of the automated systems that would inherit part of the workload.

The welding robots at Shimizu were developed with the people who understood welding, not by robotics specialists working in isolation. The craft intelligence was not discarded; it was rehoused.

Industrialisation, done well, is not the eradication of craft. It is its translation into a different medium. Done badly, it is the loss of a form of intelligence that took generations to accumulate and that no specification, no digital twin, and no training module can replace.

The cost we are not pricing

The UK faces the same transition, but with a training system still largely organised around older occupational boundaries. The Offsite Alliance has mapped 19 future occupational profiles across design, manufacture, integration and installation, reflecting the rise of digital production, platform-based delivery and building performance.

Yet funding rules, career pathways and training provision have not kept pace with either the shift in how buildings are designed and delivered or their growing technical complexity. Training also remains anchored to projects, not people, in a sector where a brickwork package may last four months and the apprentice attached to it has nowhere to go when the project ends.

Mace gives that discontinuity a price: on a 500-operative site, moving from a stable to a high-churn workforce was associated with roughly 8% lower output per hour.

We risk designing the training system around the workforce we are losing, when we should be designing it around the knowledge we cannot afford to lose. One counts roles. The other asks what forms of judgement must survive when those roles change.